Artificial Intelligence unit five notes ppt

AIML

1. APPLICATIONS OF AI
Major Applications of AI are:
Natural Language Processing
Big Data Analytics
Data Science
Robotics

2. Natural Language Processing
•Language Models –Types
1.State Machines –FSM/IFSM –D/ND
2.Rule Systems –Grammars/Syntatic/Semantic
3.Logic –Predicate/Propositional
4.Probabilistic Models –Markov Model
•Terms used in Communication
1.Speech Act –Produces Language
2.Speaker –Producer of Speech
3.Word –Communicative Sign
•Understanding
The process of mapping into appropriate actions.
•What makes Understanding Hard?
1.Complexity of the Target.
2.Types of Mapping: 1-1,1-M, M-1, M-M.
3.Level of Interaction
4.Presence of Noise.

3. •Steps in NLP:
1. Morphological Analysis: Letter->Units->Words
2. Syntactic Analysis
3. Semantic Analysis
4. Pragmatic Analysis
5. Discourse Integration
•Terminologies:
1.Formal Language
2.String
3.Grammar
4.Noun Phrases
NP, VP, Sentences : Non-Terminal Symbols
Rules : Productions.
•Steps in Communication B/w Speaker Agent & Hearer Agent
1. Intention 5. Analysis
2. Generation 6. Disambiguation
3. Synthesis 7. Incorporation
4. Perception

4. Knowledge Levels Used in NLP:
1.Phonological
2.Morphological
3.Syntactic
4.Semantic
5.Pragmatic
6.World

5. •Process of Sending documents / text / data in which the user is
interested.
•Characteristics of IR
1. A Huge Data/Document Collection -BD.
2. A format of Query with standard query language.
3. The generated Result Model.
4. Displaying Results Model.
•Evaluation of IR System
1. Precision –Propotionate of O/P to I/P
2. Recall –Propotionate of Relevant Docs in KB
ROC (Receiver Operating Characteristic) Curve
IR Refinement
Synonyms
Spelling Correction
Meta data & Result Presentation
InformationRetrieval & Extraction

6. •Classification & Clustering –UnSupervisedLearning –K
| Means Clustering
Supervised Learning –Decision Tree
Information Extraction –Technique of Creating DB Entries
•Cascade Finite Transducer –FSM
1.Tokenization --Characters to Tokens
2.Complex Words Handling --FS Grammar Rules
3.Basic Group Handling --NG/VG/CJ/PR
4.Complex Phrase Handling –Grouping of Complex Phrases
5.Structure Merging

7. Big Data & InformationRetrieval
•Big Data –Gathering Storing Large Amounts of Data
•Characteristics:
1.Volume –Size
2.Variety –Heterogenous
3.Velocity –Speed of Generation
4.Variability –Inconsistency
•Categories
1.Structured
2.Unstructured
3.Semi-Structured
•To improve efficiency we orderizeBig Data.

8. •Big Data Process
Data Management Analytics
Acquisition & Extraction, Integration, Modeling & Interpretation
Recording Cleaning & Aggregation & Analysis
Annotation Representation
BD can be: Text, Audio, Video, Social Media & Multimedia.

9. •Speech Analytics
1. Neural Networks -unstructure
2. Artificial Neural Networks
3. Machine Learning -Structral
4. Deep Learning
•Speech Recognition Components:
1.Capturing Device
2.Processor
3.Storage --Preprocessed Storage & Reference Speech Pattern
•Speech Recognition Tools:
1.Kaldi–C++
2.CMUSphinx–Java
3.Julius –C
4.HTK –C
5.Simon –C++

10. Robotics
•3 types:
1. Variable Sequence Robot
2. Playback Robot
3. Intelligent Robot
•Characteristics:
1.Machines –Mechanical Devices
2.Automatic –Automated Operations
3.Reprogrammable
4.Responsive
Terminologies:
1.Path Planning –Bug Algorithm
2.Localization for a point
3.Sensing a point
4.Mapping a point

11. •Hardwares:
1. Motion –Terrestrial, Airborne, Aquatic & Space.
Batteries & Motors –Stepper & Servo
2. Sensing
3. Reasoning
4.Communication
Most Commonly Used:
Servo Motors, Gears, Wheeled Mobile Robots, Steeled Wheels,
Complex Wheels, Trackers, Limbs, Sensors, NonVisual
Sensors, Contact Sensors, Tactile Sensors, Inertial Sensors (
Gyroscopes, Accelerometers, Magnetometers), Infra Red
Sensor.
Common Examples: SONAR, Laser RangeFinders.

12. •Architectural Models
1.Hierarchical Control
2.Reactive / BehaviourBased Paradigm
3.Hybrid Control Architecture (Middleware, High Level Control)
SPA –Sense Plan Act
1.Sense
2.Vision
3.Reasoning
4.Pilot
Horizontal Decomposition:
Composition –Model –Plan –Execute –Motor Control –Action
Sensing
Environment